Question: The following data were generated using Y = 4 + 3X + ?, where ? has a chi-square distribution with two degrees of freedom (a highly skewed distribution). Use the Atkinson simulated-envelope approach to check for nonnormality of the errors, and then use the Flack and Flores modified approach. Do this three times for each approach (remember that the envelopes will almost certainly differ because simulation is being used). Then construct a normal probability plot of the standardized residuals. Knowing that the errors do have a highly skewed distribution, which seems to be the more informative, the normal plot with or without the envelopes? Notice that you are asked to construct a normal probability plot of the standardized residuals, whereas the envelopes are constructed using standardized deletion residuals. Should it make any difference which is used, especially for this data set? Note that the Minitab macros ATKINSON and FLACFLRS can be used to generate the envelopes. See the chapter appendix for details.
